While OpenAI launches Sora and kills it, while Google experiments with Veo, and while Hollywood argues about whether AI video will replace cinematographers, China is doing what China does best: building the infrastructure to dominate the market before anyone else realizes there is a race.

ShengShu Technology just closed a 2 billion yuan ($293 million) funding round led by Alibaba Cloud, according to Reuters and Bloomberg. The company is building AI video generation tools and positioning itself as a contender in what it calls the path to artificial general intelligence.

You almost certainly have not heard of ShengShu. That is the point.

The Parallel AI Video Ecosystem

China is not trying to compete with American AI video companies. It is building a completely separate ecosystem. ShengShu is one of several Chinese startups racing to build AI-powered video generation platforms, and the funding pace is accelerating. Alibaba is not just writing checks; its cloud division is leading the round, which means ShengShu gets compute infrastructure on top of capital.

This is the same playbook we have seen with text-based AI. DeepSeek emerged from nowhere to challenge OpenAI. Alibaba's Qwen models now rival Claude in agentic benchmarks. Chinese companies consistently find ways to match Western capabilities at a fraction of the cost. There is no reason to think AI video will be any different.

Why $293 Million Is a Bigger Deal Than It Looks

In America, $293 million gets you a mid-tier Series C and a nice office in San Francisco. In China, it buys you an army of engineers, unlimited Alibaba Cloud credits, and enough compute to train models that will be deployed to a domestic market of 1.4 billion people. The purchasing power disparity is China's biggest structural advantage in the AI race, and Western analysts consistently underestimate it.

Bloomberg describes ShengShu as a young contender in China's crowded AI video contest. That phrasing tells you everything. In China, AI video generation is already a crowded market. In America, it is still a product demo from OpenAI that they discontinued.

The AGI Angle

ShengShu is framing its work as part of the path to AGI. That is partly marketing, but it is also partly real. Video generation requires models to understand physics, spatial relationships, temporal coherence, and causality in ways that text models do not. The companies that crack real-time, physics-accurate video generation will have solved many of the hardest remaining problems in AI.

If China gets there first, the implications go well beyond movies and marketing. AI video generation is the foundation for robotics simulation, autonomous vehicle training, synthetic data pipelines, and military scenario modeling. Whoever owns the best world models owns the future of physical AI.

The West is watching Sora die and Veo iterate. China is writing $293 million checks and building. The pattern should look familiar by now.

First reported by Reuters and Bloomberg.